What is every open source ai model from OpenAI
#what are OpenAI’s open source models
oblique zinc
hazy hollyBOT
Hi @oblique zinc. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
OpenAI has released several open-source repositories, but according to the knowledge sources provided, OpenAI has not open-sourced any of its large language models (such as GPT-3, GPT-4, GPT-4o, or the o-series models). Instead, OpenAI has open-sourced supporting tools and libraries, including:
- Whisper: An open-source speech-to-text model, used for transcribing and translating audio to text. This is the only AI model from OpenAI explicitly described as open source in the provided sources.
- tiktoken: A library for counting tokens, not a generative AI model.
- simple-evals: A simple evaluation library.
- mle-bench: A library to evaluate machine learning engineer agents.
- gym: A reinforcement learning library.
- swarm: An educational orchestration repository.
Of these, only Whisper is an actual AI model. The others are tools, libraries, or environments for research and development, not generative or reasoning models themselves. There is no mention in the knowledge sources of OpenAI open-sourcing any of its GPT, DALL·E, or o-series models.
Summary:
The only open-source AI model from OpenAI, according to the provided knowledge sources, is Whisper (speech-to-text). Other open-source projects are libraries and tools, not generative AI models What are voice chats? Other OpenAI repositories.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
oblique zinc
@hazy holly what about gpt3 or 2
hazy hollyBOT
Let me check the knowledge base for you again. One moment! :books:
Based on the provided knowledge sources, there is no explicit information confirming that OpenAI has open-sourced GPT-3 or GPT-2. The sources mention that OpenAI chose to release its models through an API rather than open-sourcing them, primarily for reasons related to commercial viability and safety monitoring Why did OpenAI choose to release an API instead of open-sourcing the models?.
Therefore, according to the knowledge sources, GPT-3 and GPT-2 are not open source. If you need access to these models, you must use OpenAI’s API platform.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
oblique zinc
@hazy holly what is codex
hazy hollyBOT
Let me check the knowledge base for you again. One moment! :books:
OpenAI Codex CLI is an open-source command-line tool that brings the power of OpenAI’s latest reasoning models directly to your terminal. It acts as a lightweight coding agent that can read, modify, and run code on your local machine to help you build features faster, squash bugs, and understand unfamiliar code. The CLI runs locally, so your source code never leaves your environment unless you choose to share it Overview.
By default, Codex targets the o4-mini model for fast reasoning, but you can specify any model available in the Responses API (e.g., codex -m o3). Codex supports multimodal inputs (text, screenshots, diagrams), offers a rich approvals workflow, and is designed for quick iteration entirely within your terminal Key Functionality FAQ.
For more details, you can visit the Codex GitHub repository.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
oblique zinc
@hazy holly is codex fully automated so I can just let it live in my terminal and do stuff on its own like run, debug, etc
hazy hollyBOT
Let me check the knowledge base for you again. One moment! :books:
Codex CLI offers several modes of operation, including a Full Auto mode. In Full Auto, Codex can autonomously read, write, and execute commands inside a sandboxed, network-disabled environment scoped to the current directory. This allows Codex to handle longer tasks—such as fixing a broken build or prototyping features—without requiring your constant supervision. You can, for example, let it run, debug, and iterate on code while you step away Approval Modes.
However, Codex will warn you before entering Full Auto if the directory isn’t under version control, and you should ensure the sandbox has permission to access required directories. While Full Auto is designed for autonomous operation, you still need to initiate tasks and set the appropriate mode Key Functionality.
In summary:
Yes, Codex can operate fully autonomously in Full Auto mode, running, debugging, and making changes on its own within the defined sandbox.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
oblique zinc
@hazy holly so if it does something wrong it runs and corrects?
hazy hollyBOT
Let me check the knowledge base for you again. One moment! :books:
In Full Auto mode, Codex CLI can autonomously read, write, and execute commands in a sandboxed environment. While the knowledge sources confirm that Codex can handle longer tasks like fixing a broken build or prototyping features without user intervention, they do not explicitly state whether Codex will automatically detect and correct its own mistakes if something goes wrong during execution. The documentation emphasizes autonomy within the sandbox but does not detail self-correction or iterative debugging behavior Approval Modes.